Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

Dataset: 80% C4 (Web Text) and 20% Python Code

Neuron 3072 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l')

Text #0

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0395. Min Act: -0.0001

Data Index: 955669 (C4 (Web Text))

Max Activating Token Index: 1005

Text #1

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0382. Min Act: -0.0001

Data Index: 590287 (C4 (Web Text))

Max Activating Token Index: 1008

Text #2

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0379. Min Act: -0.0001

Data Index: 1290114 (C4 (Web Text))

Max Activating Token Index: 1006

Text #3

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0370. Min Act: -0.0001

Data Index: 568619 (C4 (Web Text))

Max Activating Token Index: 919

Text #4

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0358. Min Act: -0.0001

Data Index: 722274 (C4 (Web Text))

Max Activating Token Index: 894

Text #5

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0381. Min Act: -0.0001

Data Index: 1145358 (C4 (Web Text))

Max Activating Token Index: 966

Text #6

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0359. Min Act: -0.0001

Data Index: 976189 (C4 (Web Text))

Max Activating Token Index: 994

Text #7

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0360. Min Act: -0.0001

Data Index: 1083604 (C4 (Web Text))

Max Activating Token Index: 981

Text #8

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0351. Min Act: -0.0001

Data Index: 879152 (C4 (Web Text))

Max Activating Token Index: 1019

Text #9

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0355. Min Act: -0.0001

Data Index: 576817 (C4 (Web Text))

Max Activating Token Index: 938

Text #10

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0338. Min Act: -0.0001

Data Index: 1100520 (C4 (Web Text))

Max Activating Token Index: 957

Text #11

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0345. Min Act: -0.0001

Data Index: 916181 (C4 (Web Text))

Max Activating Token Index: 956

Text #12

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0334. Min Act: -0.0001

Data Index: 1041608 (C4 (Web Text))

Max Activating Token Index: 1017

Text #13

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0342. Min Act: -0.0001

Data Index: 991982 (C4 (Web Text))

Max Activating Token Index: 974

Text #14

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0346. Min Act: -0.0001

Data Index: 1634467 (Python Code)

Max Activating Token Index: 1014

Text #15

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0348. Min Act: -0.0001

Data Index: 355046 (C4 (Web Text))

Max Activating Token Index: 899

Text #16

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0338. Min Act: -0.0001

Data Index: 1206738 (C4 (Web Text))

Max Activating Token Index: 1022

Text #17

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0348. Min Act: -0.0001

Data Index: 655869 (C4 (Web Text))

Max Activating Token Index: 889

Text #18

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0344. Min Act: -0.0001

Data Index: 63376 (C4 (Web Text))

Max Activating Token Index: 1009

Text #19

Max Range: 0.0395. Min Range: -0.0395

Max Act: 0.0336. Min Act: -0.0001

Data Index: 426705 (C4 (Web Text))

Max Activating Token Index: 931